mcp-librenms

MCP (Model Context Protocol) server that exposes the LibreNMS REST API as MCP tools. Built with FastMCP. Connect AI assistants (Claude Desktop, GitHub Copilot, etc.) to your LibreNMS instance to query and manage monitored devices, ports, alerts, logs, and more.

Features

~40 tools covering:

Category	Tools
System	ping, system_info
Devices	list_devices, get_device, add_device, delete_device, update_device_field, rename_device, discover_device, get_device_ports, get_device_ip_addresses, get_device_availability, get_device_outages, get_device_groups, get_device_components, get_device_graphs, get_device_maintenance, set_device_maintenance, add_device_eventlog
Alerts	list_alerts, get_alert, ack_alert, unmute_alert, list_alert_rules, get_alert_rule, delete_alert_rule
Ports	get_all_ports, search_ports, get_port_info, get_port_ip_info, ports_with_mac, update_port_description
Logs	list_eventlog, list_syslog, list_alertlog, list_authlog
Locations	list_locations, get_location, add_location, delete_location, edit_location
Sensors	list_sensors
Device groups	list_devicegroups, get_devicegroup
ARP	list_arp
Services	list_services
Inventory	get_inventory

Requirements

• Python 3.9+
• A LibreNMS instance with API access
• An API token (LibreNMS web UI: Settings > API > API Settings)

Setup

# 1. Create and activate a virtual environment
python -m venv .venv
.venv\Scripts\activate        # Windows
source .venv/bin/activate     # Linux/macOS

# 2. Install the package (editable)
pip install -e .

# 3. Configure
copy config.example.yaml config.yaml   # Windows
# cp config.example.yaml config.yaml   # Linux/macOS
# then edit config.yaml with your LibreNMS URL and API token

Or
pipx install "git+https://github.com/LLMNetOps/mcp-librenms.git"

Or
pipx install mcp-librenms

Configuration

config.yaml (gitignored — never commit it):

librenms:
  url: "http://your-librenms-host"   # base URL, no trailing slash
  token: "your-api-token"            # LibreNMS web UI: Settings > API > API Settings

Running

The server speaks MCP over streamable HTTP by default, bound to 0.0.0.0:5757:

# Console script (foreground)
mcp-librenms --config config.yaml

# Or as a module
python -m mcp_librenms --config config.yaml

Endpoint: http://<IP>:5757/mcp

Options

Flag	Default	Description
--config FILE	config.yaml	Path to YAML config file (CWD, then project root)
--host IP	0.0.0.0	Bind address
--port PORT	5757	Bind port
--daemon	off	Run in background (detached process)
--stdio	off	Use stdio transport instead of HTTP

Examples:

# Background on a custom address/port
mcp-librenms --config config.yaml --daemon --host 192.168.1.10 --port 8080
# → http://192.168.1.10:8080/mcp

# stdio transport (for MCP clients that spawn the server)
mcp-librenms --config config.yaml --stdio

MCP client configuration

Example for any MCP client that supports streamable HTTP servers:

{
  "mcpServers": {
    "librenms": {
      "url": "http://your-librenms-host:5757/mcp"
    }
  }
}

For stdio-based clients (Claude Desktop, etc.), spawn the server with --stdio and point --config at your config file:

{
  "mcpServers": {
    "librenms": {
      "command": "D:\\path\\to\\.venv\\Scripts\\mcp-librenms.exe",
      "args": ["--config", "D:\\path\\to\\mcp-librenms\\config.yaml", "--stdio"]
    }
  }
}

Note: the client uses verify=False for TLS, so it works with LibreNMS instances behind self-signed certificates.

Project structure

mcp-librenms/
├── config.example.yaml   # Template for local configuration
├── config.yaml           # Your local configuration (gitignored)
├── .gitignore
├── README.md
├── pyproject.toml        # Package metadata + console script
├── requirements.txt      # Flat dependency list
└── mcp_librenms/
    ├── __init__.py
    ├── __main__.py       # python -m mcp_librenms
    ├── client.py         # LibreNMS REST API HTTP client
    └── server.py         # MCP server + tool definitions

License

MIT